Prepared for

Keystone Recognition

What I found inside your shop, what I built you, and what's worth pursuing next — in your words, ready to act on.

Prepared by Nate St. Pierre
Engagement AI Enablement
Date June 16, 2026
01
What I Found

I spent two days inside your shop — across order entry, pricing, product setup, the floor, and the email queue — and recorded the conversations so I could actually listen and think rather than scribble notes. This isn't a business that's broken. It runs, and it runs on real skill. The opportunities here aren't about fixing failures; they're about taking the weight off the handful of people carrying the most, and getting your systems to stop making you do the same work twice.

So much of how it works lives in people's heads

A lot of key knowledge lives in just a few people's heads — pricing and new-product setup, estimating and scheduling judgment. When the right person is out, the work stalls, and much of that knowledge has never been written down. It's not a knock on anyone; it's just risk that's grown quietly as the business has.

The same information, entered by hand again and again

Product and pricing data gets keyed into Ordova, then reformatted for each of the two main industry distributor catalogs — every time, by hand. A purchase order comes in and someone retypes every field into Ordova. Freight quotes mean parsing the order and visiting the UPS site. None of it is hard; all of it is slow, and every re-keying is a place an error can sneak in.

A capable team, in a season of change

You're doing this through a real stretch of change, with order entry recently brought back in-house. That's exactly when the knowledge-in-people's-heads problem gets expensive. The good news: you're already comfortable with AI — Megan built a freight tool herself, Janet uses Claude daily, Sandra runs it as her go-to. So this isn't about convincing anyone the tools work. It's about pointing them at the few places that save the most time and protect the business from depending on any one person.

02
What I Recommend

The opportunities worth acting on — led by the one with the clearest dollar case, followed by two you've told me you want when the timing's right.

01
Order Intake Automation
A PO comes in and gets hand-typed into Ordova field by field. This reads the PO, finds the right prior order to carry specs forward, and pre-fills it so a person reviews instead of types. The clearest dollar case on the board — and the one with a proposal already in your hands.
Medium · Nate + Pete
$3K scope → $15–30K
02
Product Entry & Sync
Enter a product once instead of keying it into Ordova and the two distributor catalogs by hand — and post the hundreds of products (crystal especially) sitting unlisted today. The growth limiter you've named for a while.
Medium · Nate + Pete
$15–30K
03
Labor Data → Trustworthy Numbers
Turn the labor data you already collect into numbers you actually trust — to schedule, to quote, and eventually to predict job times with. A genuinely valuable one.
Small → Medium
$10–30K
03
What I Built & Delivered

What you're walking away with from this engagement — working tools and delivered sessions, not slides. Each tool has a full handoff doc.

Freight Quote Tool
Built · Nate + Pete
A web app that pulls freight quotes for an order automatically — no parsing the order by hand, keying in parts and dimensions, or visiting the UPS site. Pull the order number and quantity from a list and it returns the quotes. The finished, shareable version of the tool Megan prototyped, with credentials handled securely so the whole team can use it without the Claude extension.
Documentation: the Product Requirements, Technical Design & Implementation Plan behind this build.
Keystone Company Brain
Gift
Claude with deep Keystone context baked in — over 50 pages of your business — so anyone on the team can brainstorm, get options, or think through a decision without re-explaining the company every time. First shown in your AI productivity session; it's yours to own and grow.
Product Concept Studio
Gem · proof of concept
A proof of concept for turning a blank product photo plus a customer's logo into a realistic, in-use marketing image. It's not meant to replace your in-house design tools — for finished work, Photoshop and Adobe's AI features are the better fit (as we talked about). Where it earns its keep is fast, at-scale imagery brainstorming and mockups: sketch a concept here quickly, then take the direction into the real design tools to build.
Cycle Count Weeder
Gem
Takes Tara's raw Ordova cycle-count export and returns only the rows she'd actually count — stripping charges, freight, drop-ship, and inactive items — with a summary of what it removed. Built from her own idea and handed off for her to tune to her categories.
Reply Drafter
Gem
Drafts on-brand replies to customer emails — A-list accounts get white-glove care, everyone gets accuracy — with a human reviewing before anything sends. Built live in the room as a teaching walkthrough, with sample tier and house-voice files you swap for your own.
AI Productivity Training Day
Session
A working session with the whole team — not AI 101 — covering the tools you actually use, prompting, model choice, images, and how to build your own simple assistants, grounded in Keystone's real work.
Power-User Session
Session
A deeper session — Megan, Sandra, and Janet — on how these systems actually get built end to end. It landed: all three came out wanting to build for themselves (the follow-on is under What We're Pursuing).

Also in your package: your About-Us business profile (a client-owned narrative of the business, for bringing a new hire, vendor, or consultant up to speed), your Learn bundle (the AI-productivity field guide plus a build-your-own quick-start), your session notes from every conversation, and this Map. Everything here is yours to keep.

04
What We're Pursuing

Two next steps you've aligned on — the order-entry build (with a proposal already in your hands) and the deeper build-your-own engagement that came out of the power-user session. Here's what each one is, what it'd take, and what done looks like.

Order Intake (PO → Ordova)
Medium · Nate + Pete
The Problem
"A bunch of copying and pasting." A PO and artwork email arrives, someone looks the customer up in Ordova, clones a prior order so the production notes carry forward, and hand-keys every field. "This is the backbone of the business" — and it's a lot of manual work, now that it's back in-house.
What Gets Built
A wrap-around-Ordova web app that reads each PO from your existing shared Outlook inbox, finds the closest prior order so specs carry forward, and pre-fills the whole order. A person reviews instead of types; anything low-confidence routes to a "needs a human" lane. It writes into Ordova through Ordova's official API and runs in your own Cloudflare account — you own it.
What You Bring
A signed Ordova API agreement (no fee) and credential; Ordova's answer on whether the API can create an order header; 3–5 sample POs from different distributors; one hour shadowing an operator; a short field-mapping workshop; and access to the shared orders inbox.
What Done Looks Like
A standard PO arrives and lands in Ordova as a reviewed, approved, correctly-structured order in a few minutes of review instead of full hand-keying — consistent no matter who's on duty. The $3K scoping step comes first: a PRD, technical design, plan, and firm quote, with the riskiest pieces proven — and it credits toward the build.
$3K scoping → $15–30K build
Scoping 1–2 wks · build 2–4 wks
Forge Tier 1: Build Your Own
Small · Nate
The Problem
All three of you — Megan, Sandra, and Janet — came out of the power-user session wanting to build these systems yourselves, not just receive them. "It's tantalizing… I'm inclined to learn more about doing it ourselves — it gives us the most flexibility and options." The appetite is real; what's missing is the runway to get from interested to self-sufficient without months of solo trial and error.
What Gets Built
A hands-on engagement where I set up your environment — the development tool, Claude Code, and a business Claude account — then build your first real working systems with you, starting from workflows you each own (Megan's meeting-to-follow-ups flow is a natural first one). We build live, narrating every step so it's repeatable. What you come out with isn't a finished product handed over — it's a working setup you own and the know-how to keep building on your own.
What You Bring
A business Claude account on the ~$200/mo plan — the higher tier, sized for the three of you as active build-heavy users (the free and entry plans won't carry this kind of work); the people who want to learn (Megan, Sandra, and Janet); a few of your own real workflows to build against; and the time to sit in the morning sessions and practice between them.
What Done Looks Like
Each participant has built at least one working tool for their own job during the engagement and can scaffold a new one on their own afterward — Keystone can keep extending its own systems without needing me in the room.
$10,000
~1 week (mornings) · remote
05
Full Opportunity Inventory

Every opportunity identified during the engagement — the complete catalog, yours regardless of what you pursue. Built and delivered first, then the actively-pursued build, then everything on the map.

Freight Quote Tool
Parsing each order and pulling UPS quotes by hand was a four-minute hassle on every project. A secure, shareable web app now returns freight quotes from just an order number and quantity — no manual parsing, no UPS-site lookups.
Built Quick-win build
Effort
Impact
Cycle Count Weeder
Tara used to delete the non-counted products from an Ordova cycle-count export by hand. This Gem returns only the rows she'd actually count — stripping charges, freight, drop-ship, and inactive items — with a summary of what it removed.
Built Quick Win
Effort
Impact
Product / Marketing Imagery
A proof of concept that composites a blank product plus a customer's logo into a realistic in-use image — for quick imagery brainstorming and mockups at scale, before finished work moves into your in-house design tools (Photoshop / Adobe AI).
Built Proof of concept
Effort
Impact
Forge Tier 1: Build Your Own
The power-user session was delivered (Megan, Sandra, Janet) and landed — all three want to learn to build their own systems. Forge Tier 1 is the follow-on: Nate builds real systems with you until you can build your own. (Full brief above.)
Pursuing Session delivered · Small ($10K)
Effort
Impact
Order Intake (PO → Ordova)
The backbone of the business, today done entirely by hand: a PO arrives and every field is retyped into Ordova. The build reads the PO, finds the right prior order to carry specs forward, and pre-fills it so a person reviews instead of types. (Full brief above.)
Pursuing Medium
Effort
Impact
Product Entry & Sync
Product and pricing data gets keyed into Ordova, then reformatted for each of the two industry distributor catalogs — every time, by hand. Enter a product once and push to all three, plus a batch run to finally post the hundreds of unlisted products (crystal especially). The growth limiter you've named for a while.
On Map Medium
Effort
Impact
Labor Data → Trustworthy Numbers
Turn the labor data you already collect into a trusted master product list (pieces per hour, cost) you can schedule and quote on — and, at the top end, predict job times. A genuinely valuable one.
On Map Small → Large
Effort
Impact
Knowledge Capture / Continuity
So much of how Keystone runs lives in people's heads. Record someone walking through a real process once, then turn it into clear SOPs anyone can follow — continuity insurance against the risk of critical know-how living in just one person's head.
On Map Quick Win → Medium
Effort
Impact
Machine Maintenance System
Troubleshooting answers pulled from the machine manuals plus standardized maintenance logs, starting on one machine and scaling to a full floor app with schedules, history, and how-to videos. A machine down for a day is a big problem; this shortens it.
On Map Quick Win → Large
Effort
Impact
Email Draft Assistant
Reads an incoming customer email and returns the category, the right template, and a ready-to-edit draft — a person always reviews. A Reply Drafter prototype was built live in your session; the full build watches the inbox and drafts into Outlook automatically.
On Map Small
Effort
Impact
Vendor Price Change Monitor
When a vendor's prices move and yours don't follow, margin quietly erodes. Drop in a vendor price sheet plus your Ordova pricing and get back a severity-sorted report of exactly what changed, at a threshold you set. A clear dollar case.
On Map Small
Effort
Impact
Production Scheduling Assistant
Tara sets the next day's assignments from a mental model. A tool that suggests a production sequence with flags — most valuable once the labor data matures into numbers you trust.
On Map Quick Win → Small
Effort
Impact
Recurring Order Specs
About 80% of orders repeat each year, and keeping every spec consistent is a note-taking challenge. A per-client program spec card so repeat orders aren't rebuilt from memory. (Largely handled today by order management; lower appetite.)
On Map Small
Effort
Impact
11th-Month Anniversary Outreach
You reach out to customers on their 11-month anniversary — manual and dependent on remembering. Auto-surface orders hitting 11 months and draft the outreach for review. Bundles into a larger Microsoft-stack automation.
On Map Small
Effort
Impact
06
What's Next

Four ways forward, no pressure.

Path A
Continue with Nate
The smaller, high-value builds — the Vendor Price Monitor, the Email Draft Assistant, or Knowledge Capture SOPs. This is also where Forge Tier 1: Build Your Own lives — Megan, Sandra, and Janet all wanted it: a focused engagement where I build a few real systems with you and show you how, so you come out able to build a lot more on your own ($10,000).
$10,000–$15,000 · Nate solo · 2–4 weeks
Path B
Nate + Pete Build
For the builds that need real engineering alongside the AI — Order Intake (the one with a proposal in your hands), Product Entry & Sync, the Labor Data system, or the Machine Maintenance step-up. Pete builds the backend and system connections; I build the intelligence and experience on top.
$15,000–$30,000 · Nate + Pete · 2–4 weeks
Path C
Bring in Rogue Agents
For the most ambitious version — the full machine-maintenance floor app, or a complete labor and scheduling system at scale. Full team: Nate (creative technology + AI), Pete (engineering), Sarah (account + delivery). I facilitate the intro and stay involved.
$50,000+ · Full team · 4–6 weeks
Path D
Take the Map Forward
Everything here is specific enough to execute without us — run it with your own team, or hand it to another developer. The pursuing brief and the full inventory are written to be briefable. This Map is yours either way.
Full handoff complete · the map is yours to keep